67°F

Aaron Parecki

  • Articles
  • Notes
  • Photos
  • David Waite https://twitter.com/dwaite   •   Jun 6
    @aaronpk Most surprisingly interesting thing I got from https://developer.okta.com/blog/2019/06/04/what-the-heck-is-sign-in-with-apple - you managed to get http://example-app.com
    Aaron Parecki
    ๐Ÿ˜‚๐Ÿ˜‚๐Ÿ˜‚ Yep I use it in my book and I've also been using it for testing out redirect URIs in workshops and stuff!
    Portland, Oregon
    1 reply
    Thu, Jun 6, 2019 4:46pm -07:00
  • Wizages - Sam Patzer https://twitter.com/Wizages   •   Jun 6
    Just for the uniqueid. Verify that the uniqueid is not modified.
    Aaron Parecki
    verify where? The unique ID comes back in the ID token not the access token. (also happy to take this to DM)
    Portland, Oregon
    Thu, Jun 6, 2019 4:20pm -07:00
  • Wizages - Sam Patzer https://twitter.com/Wizages   •   Jun 6
    So talked with the Apple engineers here at WWDC:
    They don't have that endpoint, they also will not expose user_info or a revocation endpoint. The user_info will only be sent once and only once then you will only get a unique id again. Only scopes available now are name and email
    Aaron Parecki
    Another question, if there is no `user_info` endpoint, what are the access token and refresh tokens for?
    Portland, Oregon
    1 reply
    Thu, Jun 6, 2019 4:18pm -07:00
  • Wizages - Sam Patzer https://twitter.com/Wizages   •   Jun 6
    Yep. That seems like a bug. If you forget it well you are left without it. You should be able to request new scopes...
    Aaron Parecki
    Progress! I now get the screen which lets me edit my name and choose the email to share. I only see that the first time, all subsequent requests show a confirmation only.

    Still no luck actually getting the email address back in the ID token though.
    Portland, Oregon
    1 like
    Thu, Jun 6, 2019 4:07pm -07:00
  • Wizages - Sam Patzer https://twitter.com/Wizages   •   Jun 6
    That is not a bug that is feature. They told me they will only give you the info once. Probably why scope wonโ€™t matter after your first invoke.
    Aaron Parecki
    I will go test this out with new app credentials though to confirm. Thanks for the lead!
    Portland, Oregon
    1 like
    Thu, Jun 6, 2019 3:28pm -07:00
  • Wizages - Sam Patzer https://twitter.com/Wizages   •   Jun 6
    That is not a bug that is feature. They told me they will only give you the info once. Probably why scope wonโ€™t matter after your first invoke.
    Aaron Parecki
    interesting. well the bug is that I have *never* gotten it, because I didn't request it the first time, and now I can't request it ever again.
    Portland, Oregon
    1 reply
    Thu, Jun 6, 2019 3:28pm -07:00
  • Wizages - Sam Patzer https://twitter.com/Wizages   •   Jun 6
    So talked with the Apple engineers here at WWDC:
    They don't have that endpoint, they also will not expose user_info or a revocation endpoint. The user_info will only be sent once and only once then you will only get a unique id again. Only scopes available now are name and email
    Aaron Parecki
    Just verified again, and I don't get back name or email address when I request "name email" scope.

    I did find a bug where apparently Apple is ignoring the "scope" parameter after the very first time you authorize an app though, so could be related.
    Portland, Oregon
    3 replies
    Thu, Jun 6, 2019 2:26pm -07:00
  • Wizages - Sam Patzer https://twitter.com/Wizages   •   Jun 6
    So talked with the Apple engineers here at WWDC:
    They don't have that endpoint, they also will not expose user_info or a revocation endpoint. The user_info will only be sent once and only once then you will only get a unique id again. Only scopes available now are name and email
    Aaron Parecki
    Brilliant, thanks for the info!

    Have you been able to successfully request name and email scope yet? It wasn't working in my testing.
    Portland, Oregon
    Thu, Jun 6, 2019 2:22pm -07:00
  • mefarazath https://github.com/mefarazath   •   Jun 6

    #1 Having 'scope' parameter in the authorization request seems to cause login failures

    Aaron Parecki
    Thank you! You helped me track down a deeper problem!

    It turns out that you're right, having `scope` in the request causes Apple to return a 500 server error when confirming the login on their site before it redirects back to the app.

    **However**, the really weird part is that Apple apparently completely ignores the `scope` parameter the second time you log in to an app, so there is no error.

    The very first time I logged in to an app while testing this code I didn't include the `scope` parameter, so it worked. Then I added the parameter to see if I could get it to return an email address, and it didn't. It also didn't fail, because I had already logged in once.

    I was having trouble logging in with a new App ID I created, and this is the reason! I just tried removing the scope from my attempt and now I'm able to log in with new App IDs.

    This is very inconsistent behavior by Apple, so I hope they fix it later.
    Portland, Oregon, USA
    Thu, Jun 6, 2019 12:29pm -07:00
  • Tim Ysewyn @ ๐Ÿ‡ง๐Ÿ‡ช๐Ÿ  https://twitter.com/TYsewyn   •   Jun 6
    Why should the role be in the token if you have the userinfo endpoint? Or why should there even be a (list of) role(s) in the token if itโ€™s only a means to have access to an endpoint? ๐Ÿค”
    Aaron Parecki
    Some people like to use JWTs for access tokens or other self-encoded mechanisms. There are definitely trade-offs.
    Portland, Oregon
    Thu, Jun 6, 2019 12:20pm -07:00
  • https://rhiaro.co.uk/ https://toot.cat/@rhiaro   •   Jun 6

    I once bought into the "if you don't post to your own website you don't have anything valuable to offer on technology" theory but eventually realised there are other things in tech and, shock horror, even the Web, than posting on your own website. I still see it being thrown around as an excuse to dismiss out of hand work that people don't understand immediately and find directly relevant to their specific needs though.

    Posting this on my own website, so it must be true. https://rhiaro.co.uk/2019/06/theory

    Aaron Parecki
    echoing the rest of that conversation, I think the larger point is the inverse.

    If you *only* post your thoughts about decentralization on platforms that actively work against decentralization and cause real-world harm to people, then maybe it's okay to ignore those thoughts.
    Portland, Oregon
    Thu, Jun 6, 2019 12:17pm -07:00
  • Stephan https://twitter.com/Stephan007   •   Jun 6
    Nice write up! Question: if an authenticated user gets a new/extra role, does the server create a new JWT or is there a way to update the existing token?
    Aaron Parecki
    if your access tokens are just a reference to a record in a database (the hotel key is just a number, and the doors look up access info in a central server), then you can update the roles in the existing token.
    Portland, Oregon
    Thu, Jun 6, 2019 12:13pm -07:00
  • Stephan https://twitter.com/Stephan007   •   Jun 6
    Nice write up! Question: if an authenticated user gets a new/extra role, does the server create a new JWT or is there a way to update the existing token?
    Aaron Parecki
    The analogy continues... with JWT access tokens, that's like encoding access data into the hotel key card. You'd have to go back to the front desk to get a new card.
    Portland, Oregon
    Thu, Jun 6, 2019 12:12pm -07:00
  • Stephan https://twitter.com/Stephan007   •   Jun 6
    Nice write up! Question: if an authenticated user gets a new/extra role, does the server create a new JWT or is there a way to update the existing token?
    Aaron Parecki
    The answer is it depends on how your access tokens / hotel key cards are implemented!
    Portland, Oregon
    Thu, Jun 6, 2019 12:11pm -07:00
  • Wizages - Sam Patzer https://twitter.com/Wizages   •   Jun 6
    Do you know where you can find the .well-known/openid-configuration on the apple url?
    Do they even use it?
    Aaron Parecki
    I haven't found it yet. I wouldn't be surprised if they just don't have that endpoint
    Portland, Oregon
    1 like 4 replies
    Thu, Jun 6, 2019 10:46am -07:00
  • Gokul Thirumalai https://twitter.com/gokult   •   Jun 5
    Thanks Alex. Where do you see the biggest gap that will help you push forward?
    Aaron Parecki
    The docs are also wrong in a few places. Happy to provide details via DM or email.
    Portland, Oregon, USA
    Wed, Jun 5, 2019 4:45pm -07:00
  • Gokul Thirumalai https://twitter.com/gokult   •   Jun 5
    Thanks Alex. Where do you see the biggest gap that will help you push forward?
    Aaron Parecki
    I have so many questions for you! I was able to create a proof of concept, but would love to know some of the missing details. The current documentation is not complete enough to make a working app, I had to guess things based on my knowledge of OIDC.

    https://developer.okta.com/blog/2019/06/04/what-the-heck-is-sign-in-with-apple
    Portland, Oregon, USA
    1 reply
    Wed, Jun 5, 2019 4:40pm -07:00
  • Nate Angell https://twitter.com/xolotl   •   Jun 5
    Lawyer at @indiewebcamp: "Why isn't anyone at my GDPR unconference session?"
    Aaron Parecki
    you joke, but: https://indieweb.org/2018/D%C3%BCsseldorf/gdpr
    Portland, Oregon, USA
    2 likes 1 repost 2 replies
    Wed, Jun 5, 2019 12:41pm -07:00
  • Simon Rice https://twitter.com/_SimonRice   •   Jun 5
    Excellent thread - just to clarify with another example since Google do identity & calendar - if my app needs OAuth to (say) read playlist data specifically from Spotify via their dev API & do nothing whatsoever user ID related with them, I donโ€™t need โ€œSign In With Appleโ€?
    Aaron Parecki
    Yes that is my understanding reading their guidelines. Of course this remains to be seen how it will play out in practice.
    Portland, Oregon
    2 likes 1 reply
    Wed, Jun 5, 2019 10:33am -07:00
  • AngeloStavrow https://micro.blog/AngeloStavrow   •   Jun 5

    @aaronpk This is a great resource, thanks! Is sign-in still working for you, though? I'm seeing the Apple sign-in page throw an error in my testing after I authenticate (though it's possible there's an error in the setup).

    Aaron Parecki
    It's working with the first app I registered in the portal, but hasn't worked with new app IDs I've made since! I'm guessing some weird Apple but that they'll probably fix soon. This is all clearly very early beta right now.
    Portland, Oregon
    1 reply
    Wed, Jun 5, 2019 9:14am -07:00
older

Hi, I'm Aaron Parecki, Director of Identity Standards at Okta, and co-founder of IndieWebCamp. I maintain oauth.net, write and consult about OAuth, and participate in the OAuth Working Group at the IETF. I also help people learn about video production and livestreaming. (detailed bio)

I've been tracking my location since 2008 and I wrote 100 songs in 100 days. I've spoken at conferences around the world about owning your data, OAuth, quantified self, and explained why R is a vowel. Read more.

  • Director of Identity Standards at Okta
  • IndieWebCamp Founder
  • OAuth WG Editor
  • OpenID Board Member

  • ๐ŸŽฅ YouTube Tutorials and Reviews
  • ๐Ÿ  We're building a triplex!
  • โญ๏ธ Life Stack
  • โš™๏ธ Home Automation
  • All
  • Articles
  • Bookmarks
  • Notes
  • Photos
  • Replies
  • Reviews
  • Trips
  • Videos
  • Contact
© 1999-2025 by Aaron Parecki. Powered by p3k. This site supports Webmention.
Except where otherwise noted, text content on this site is licensed under a Creative Commons Attribution 3.0 License.
IndieWebCamp Microformats Webmention W3C HTML5 Creative Commons
WeChat ID
aaronpk_tv